Supplementary Figure 1 – Validation of genes sensitive to p23 levels in a different MCF-7 cell line overexpressing p23 [clone #8]. qRT-PCR experiments were performed to show the relative fold-change (RFC) in the expression of six genes from Figure 1c between
MCF-7 control (white bars) and MCF-7+p23 (#8) cells (black bars). Data are means of three independent experiments, normalized to
GAPDH, and presented as RFC from that of MCF-7 control cells, set to 1. Error bars represent standard deviation.
Supplementary Figure 2 – Genes most upregulated or downregulated by p23 overexpression are also overexpressed or under- expressed in invasive breast cancers
Differential analyses of microarray datasets from invasive breast cancer subgroups (listed in the Legend) were performed using the
Oncomine database (www.oncomine.org) to determine if the ten genes most upregulated or the eighteen genes most downregulated ( >
5-fold, Tables S5-6) by p23 overexpression are also over- or under-expressed in invasive breast cancers. Shown are five of the 10 upregulated and 11 of the 18 downregulated genes that are significantly over- or under-expressed in at least one invasive breast cancer dataset (p-value < 0.05) (* designates those genes that are not considered significant using Oncomine differential analysis, but are associated with invasive cancers in referenced literature). Over-expressed genes ranked highest, based on significance, within a particular dataset are designated with red boxes (or dark blue boxes for under-expressed genes). Pink and light blue boxes indicate genes that are statistically significant, but with lower gene rank. White boxes designate genes that are not statistically significant within a given dataset. Grey boxes indicate that gene expression was not measured for a particular dataset. Supplementary Figure 3 – p23 mediated changes in gene expression upon inhibition of Hsp90
A) MCF-7+p23 cells were treated with either DMSO vehicle or 1μM GA for two hours prior to protein extraction. p23-Hsp90 co- immunoprecipitation experiments were performed following cell lysis [50mM Hepes, pH 7.6, 150mM NaCl, 1mM EDTA, 1mM
EGTA, 1mM NaF, 1% Triton X-100, and 10% glycerol, 3mM DTT, 1mM PMSF, 5mM ATP, 20mM sodium molybdinate and protease inhibitor cocktail (Sigma)] and pre- cleared with protein-G agarose beads (Roche). Protein lysate (8mg) was incubated overnight at 4°C with 10μg HA.11 antibody (Covance) and 0.2% input and 1/5 of the IP reaction was immunoblotted with anti-Hsp90
(1:1000; Transduction Laboratories, H38220) or p23. B) MCF-7+p23 cells were hormone-starved for three days and treated for 48 hours with GA (left panel) or 24 hours with 17-AAG (right panel). RNA was extracted and qRT-PCR performed. Data are means, normalized to 28S (left panel) or GAPDH (right panel), and expressed as relative fold-change (RFC) in gene expression to vehicle treated MCF-7+p23 cells, set to 1. Error bars represent standard deviation.
Supplementary Figure 4 – p23 levels can influence the estrogen sensitivity of some, but not all genes
Venn diagrams show the distribution of estrogen sensitivity of 357 genes identified in control and MCF-7+p23 cells. 116 were sensitive to E2 in both cell types (green overlap; 47 induced, 69 repressed). There were 51 induced and 47 repressed genes in control cells (yellow circles) that were not induced in MCF-7+p23 cells, suggesting these genes were made unresponsive to E2 by p23 overexpression. Conversely, 65 genes were induced and 78 repressed by E2 only in MCF-7+p23 cells, suggesting that p23 overexpression makes these genes more responsive to E2 (blue circles). When examined closely, the majority of the 65 genes that were uniquely induced or downregulated by E2 in the MCF-7+p23 cells displayed between a 1.2 and 1.9-fold change in mRNA expression by E2 in control cells and greater than 2-fold change in MCF-7+p23 cells (data not shown). There were also a small number of genes (11) not regulated by E2 in control cells (fold change < 1.2), but induced (or repressed) 2-fold or more by E2 upon p23 overexpression. Therefore, these genes can be classified as being highly reactive to E2 by p23 overexpression and are listed in
Supplementary Table 7.
Supplementary Figure 5 – Semi-quantitative analysis of ER recruitment to PMP22 and ABCC3 EREs
A schematic representation of the PMP22 and ABCC3 loci. Arrows represent transcriptional start sites (TSS), white boxes represent exons, the stripped boxes represent the ER binding elements (ERE) as determined by global ChIP-Chip analysis and black boxes represent the control upstream sequences (UPS). ChIP assays were preformed in MCF-7 control and MCF-7+p23 cells in the absence
(ethanol vehicle treated) and presence of E2 using an antibody against ER or equivalent amount of non-specific rabbit IgG (Sigma).
Relative levels of ER binding to 0.1% and 0.05% of the total input are shown by PCR and are representative of at least three independent experiments.
Supplementary Figure 6 - p23 expression in human breast cancer
A breast cancer tissue microarray containing 213 specimens was stained by immunohistochemistry with anti-p23 antibody and the intensity of p23 staining in the nucleus and cytoplasm was scored on a scale of 0 to 3+, with 3+ being the highest intensity staining.
Example specimens scored as 2+ (left panel) and 3+ (right panel) for cytoplasmic p23 from biopsies are shown. Insets magnify a portion of each image.
Supplementary Figure 7 – Subcellular distribution of p23 in MCF-7 control and MCF- 7+p23 cells as a function of ER agonist and antagonist treatment
Nuclear and cytoplasmic fractions were prepared using Pierce Biotechnology’s NE-PER Nuclear and Cytoplasmic Extraction
Reagents (78833, Pierce Biotechnology) according to manufacturer’s instructions, from 16h ethanol vehicle (-), 1nM 17- -estradiol
(E), 100nM 4-OH-tamoxifen (T), or 100nM ICI 182,780 (I) treated MCF-7 control and MCF-7+p23 cells. Levels of p23, TFII-I, and